Added unreachable macro to alleged unreachable code rather than using hidden default value

[Sheepyhead]

Objective

Fixes #168

Context

This doesn't seem to panic any tests, but this still confers a risk to users that this code has actually been reached by some edge cases, where it would simply fail silently before.

Feedback wanted

• If you have code depending on taffy, I'd really like to know if you get a panic from this PR, the more real dirty user code we test this on the better

[mockersf]

• If you have code depending on taffy, I'd really like to know if you get a panic from this PR, the more real dirty user code we test this on the better

No need to worry, it's impossible to fall in that case... as long as align_self() doesn't change

taffy/src/style.rs

Lines 475 to 487 in e75a55f

	pub(crate) fn align_self(&self, parent: &FlexboxLayout) -> AlignSelf {
	if self.align_self == AlignSelf::Auto {
	match parent.align_items {
	AlignItems::FlexStart => AlignSelf::FlexStart,
	AlignItems::FlexEnd => AlignSelf::FlexEnd,
	AlignItems::Center => AlignSelf::Center,
	AlignItems::Baseline => AlignSelf::Baseline,
	AlignItems::Stretch => AlignSelf::Stretch,
	}
	} else {
	self.align_self
	}
	}

It could be interesting to add a comment on that method to say it should never return AlignSelf::Auto as that's an invariant expected by other part of the code

[TimJentzsch]

This doesn't seem to panic any tests, but this still confers a risk to users that this code has actually been reached by some edge cases, where it would simply fail silently before.

IMO this is good, it's usually better to fail early and detect these kind of bugs instead of failing silently and causing problems in other areas that are hard to track back.

[Sheepyhead]

IMO this is good, it's usually better to fail early and detect these kind of bugs instead of failing silently and causing problems in other areas that are hard to track back.

I agree, still it's something to note about this innocent-looking two-line PR ;)

[Sheepyhead]

• If you have code depending on taffy, I'd really like to know if you get a panic from this PR, the more real dirty user code we test this on the better

No need to worry, it's impossible to fall in that case... as long as align_self() doesn't change

taffy/src/style.rs

Lines 475 to 487 in e75a55f

	pub(crate) fn align_self(&self, parent: &FlexboxLayout) -> AlignSelf {
	if self.align_self == AlignSelf::Auto {
	match parent.align_items {
	AlignItems::FlexStart => AlignSelf::FlexStart,
	AlignItems::FlexEnd => AlignSelf::FlexEnd,
	AlignItems::Center => AlignSelf::Center,
	AlignItems::Baseline => AlignSelf::Baseline,
	AlignItems::Stretch => AlignSelf::Stretch,
	}
	} else {
	self.align_self
	}
	}

It could be interesting to add a comment on that method to say it should never return AlignSelf::Auto as that's an invariant expected by other part of the code

Should I add that as part of this PR?

[alice-i-cecile]

Should I add that as part of this PR?

Yes please!

[Sheepyhead]

Should I add that as part of this PR?

Yes please!

Idk if the way I did it works, but it was the best I could think of

[Weibye]

Looks good :)

[Weibye]

@Sheepyhead Please run cargo fmt on the code then push the changes :) Everything else is good to go 🚀